Summary: Long-term stability of social status is the norm, so rapid assimilation of migrants is not the default expectation. The deep roots of the evidence suggest that "development follows people, not places," meaning that migrants carry enduring characteristics across borders. Standard measures of one-generation mobility overestimate actual mobility, and multigenerational data indicate much greater stability than parent-child correlations suggest. Measurement errors and poor correlations make societies appear more mobile and more assimilated than they really are. Persistence in social status has often survived huge shocks (revolutions, abolition of slavery, destruction of property, lotteries), suggesting that environmental changes cannot reliably close the gap between groups. Between 1850 and 1920, during mass European immigration, 25-40% (and sometimes as many as 60-75%) of migrants returned to their homeland, so "successful assimilation" is partly the result of survival bias. Returning migrants were negatively selected, meaning that those least successful were more likely to leave. Data on the links between grandparents and grandchildren show that differences in occupational income between ethnic groups in Europe persisted to a significant extent until 1940, contradicting the "rapid cultural melting pot" theory. Immigrants changed American culture and institutions, not just assimilated them, so assimilation is not a one-way process. Counties in the United States whose residents hail from wealthier European countries still have higher GDP per capita, suggesting a long-term impact of origin on the economy. High-achieving modern immigrant groups (e.g., Indian and East Asian) are successful mainly because they are subjected to very positive selection before arriving in America, not because America "assimilated" them into success. If assimilation were strong, above-average-performing groups would quickly approach the average, but the article gives no indication that Americans of Asian descent are about to lose their edge. Mexican immigrants are described as not undergoing high positive selection and therefore performing worse than non-Latino whites in terms of income and education. Data from the early 20th century show that Mexican immigrants were extreme outliers in terms of over-representation in prison/serious crimes compared to the Irish and Italians, undermining the "Mexicans will follow in the footsteps of the Irish/Italians" analogy. For Hispanics, the greatest improvements occur between the first and second generations, while the gains of later generations are smaller, less systematic, and can stall or reverse (including worse outcomes for third-generation men in some data). Multigenerational observation of the descendants of Mexican immigrants shows no improvement in educational attainment between generations, and sometimes even a decline, with the fourth generation's performance the furthest from that of whites. The persistence of large disparities between blacks and whites (and Native Americans) after many generations in the same country is used as evidence that America is not integrating groups in a way that ensures equality.